There is strong evidence that the aetiology of insulin-dependent diabetes mellitus (IDDM) is due to a complex interaction between genes and the environment and that the pathogenesis is autoimmune. In early perinatal life the immune system is induceable and exposures in this period may initiate autoimmunity. Recent findings of time and space clustering of birth dates for later diabetic cases together with the early observation of a very high prevalence of diabetes in cases with rubella embryopathy suggest that foetal virus exposure may be important. Recent findings from Sweden and Finland suggest that enterovirus exposure during foetal life may initiate autoimmunity which may lead to diabetes. Other immune events, such as maternal-foetal blood group incompatibility and pre-eclampsia in the mother have also been associated with IDDM risk. Other more unspecific events in the perinatal period, such as a short gestational age, caesarean section and neonatal respiratory disease, are also indicated to increase the risk. In addition, food components such as nitrosamine components, cow's milk protein and gliadin have been proposed to initiate the slowly progressing autoimmune beta-cell destruction. Most of these epidemiological findings are supported by experimental studies in the nonobese diabetic mice but their exact mechanisms of action are still unclear. It is concluded that new evidence is accumulating indicating that perinatal exposures may be important for the initiation of beta-cell destruction. As such risk factors may be targets for primary prevention strategies further studies are urgently warranted.